Pentagon Slams OpenAI Executive's Claims on Chinese AI Curb, Citing a 2026 Law: 'Not Some Deep State Scheme'

The Pentagon has expressed disapproval of OpenAI‘s Dean Ball‘s stance on regulation, raising questions about the future of OpenAI’s relationship with the government.

On Sunday, Dean Ball, head of strategic futures at OpenAI, had indicated on X that several U.S. government agencies, including the Departments of Defense, Transportation, Energy, Agriculture, Commerce, NASA, and Congress, have barred employees from using Chinese AI tools, arguing the restrictions are based on poorly justified safety concerns and have discouraged regulated companies from adopting the technology.

Defence Under Secretary Emil Michael fired back on X and refuted Ball’s assertion that government agencies, including the Pentagon, were preventing employees from using Chinese AI. 

He elaborated that Congress passed a 2026 law restricting certain uses of only DeepSeek and High Flyer models, while allowing waivers in specific cases. The measure was approved through the legislative process, not through a “Deep State” initiative, as Ball alleged.

“It went through the democratic process not some Deep State scheme like he would prefer,” Michael wrote.

Every industry/ecosystem has its supreme village idiot. @deanwball is that for AI. Congress passed a law in 2026 that restricted some uses of DeepSeek/High Flyer with waivers permitted. Only those models. It went through the democratic process not some Deep State scheme like he… https://t.co/PtPo23tqLA

— Under Secretary of War Emil Michael (@USWREMichael) July 19, 2026

Meanwhile, Former White House AI and crypto czar David Sacks criticized proposals to use regulatory uncertainty to discourage adoption of Chinese AI models, arguing that AI policy should be based on evidence and open competition. He also responded to Dean Ball, who suggested the government could steer companies away from Chinese AI by having agencies issue informal warnings that create fear, uncertainty, and doubt (FUD), rather than imposing an outright ban.

AI Access Battle Intensifies

This development comes as the Trump administration is reportedly seeking greater control over who gets early access to advanced AI models from companies like OpenAI and Anthropic. Under its new Gold Eagle cybersecurity initiative, the White House could approve which organizations participate in trusted-partner programs, shifting authority away from AI companies’ current self-managed access process.

Earlier this month, OpenAI was reportedly considering offering a 5% equity stake to the U.S. government as part of a broader effort to strengthen ties with the Trump administration and secure financial backing.

Meanwhile, Chinese AI models have overtaken U.S. rivals on OpenRouter for the first time, accounting for a record 58% of tokens processed by U.S. firms, according to data shared by The Kobeissi Letter. Their share has surged from less than 10% at the start of 2025 to as high as 63% in early July, driven by growing adoption of models from companies such as DeepSeek. OpenRouter’s token usage is widely viewed as a key indicator of real-world AI model adoption.

Disclaimer: This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors.

Image via Shutterstock